Petra Zeidler - 2026-05-19 09:47:10
so the smallest routable prefix in ILNP is a /32, do I get that right?
Jen Linkova - 2026-05-19 09:52:37
I might be missing smth but I do not think it impacts routing
Jen Linkova - 2026-05-19 09:52:50
It's all about hosts
Petra Zeidler - 2026-05-19 09:53:44
how is your host able to pick a random lower /32 bits if these are not in the same network? how does their router find them?
Marco d'Itri - 2026-05-19 09:54:51
Also: what about the impact on NDP caches?
Petra Zeidler - 2026-05-19 09:56:12
ignore me, /64 not /32
Jerry Lundström - 2026-05-19 09:57:32
that's never gonna happen then :)
Jen Linkova - 2026-05-19 09:59:20
@marco if the host doesn't geverate a new address just for this and just changed networks - none, I guess. If the approach is to generate new address in the same network - yes, indeed, it will inflate nd cache, unless /64 per host is used
Jen Linkova - 2026-05-19 10:00:18
I have some concerns re:deployability due to "beneficial middle boxes"
Éric Vyncke - 2026-05-19 10:02:33
another use of the /64 per host
Petra Zeidler - 2026-05-19 10:02:51
I have the suspicion that trackers already assume that everything in a /48 is the same ID
Éric Vyncke - 2026-05-19 10:04:02
Using the least significant 64 bits is both smart, but I am unsure about the privacy here...
Jen Linkova - 2026-05-19 10:04:41
@eric yes, that was my comment. However if the observer knows that the given network uses /64 per host, they would be looking for the network id.
Éric Vyncke - 2026-05-19 10:25:53
Nice to see an ending about IPv6
Marco d'Itri - 2026-05-19 10:31:48
as usual, the issue is "do not peer with badly run route servers at badly run IXPs", nothing fundamental here about RS
Rinse Kloek - 2026-05-19 10:44:54
Agree
Dejan Stojanović - 2026-05-19 10:50:55
it is ok